Food sources of dietary fibre and risk of total knee replacement related to severe osteoarthritis, the Singapore Chinese Health Study

Page created by Norman Adams
 
CONTINUE READING
Osteoarthritis

                                                                                                                                                                 RMD Open: first published as 10.1136/rmdopen-2021-001602 on 30 July 2021. Downloaded from http://rmdopen.bmj.com/ on October 11, 2021 by guest. Protected by copyright.
                                  ORIGINAL RESEARCH

                                  Food sources of dietary fibre and risk of
                                  total knee replacement related to severe
                                  osteoarthritis, the Singapore Chinese
                                  Health Study
                                  Ying Ying Leung ‍ ‍,1,2 Aizhen Jin,3 Kelvin Bryan Tan,4 Li-­Wei Ang,4,5
                                  Jian-­Min Yuan,6,7 Woon-­Puay Koh3

To cite: Leung YY, Jin A,         ABSTRACT
Tan KB, et al. Food sources of      Objectives We aimed to evaluate the association between
                                                                                                     Key messages
dietary fibre and risk of total     fibre intake and its food sources, and the risk of total knee
knee replacement related
                                    replacement (TKR) due to severe knee osteoarthritis (KOA).
                                                                                                     What is already known about this subject?
to severe osteoarthritis,                                                                            ►► The association between diet and development of
                                    Methods We used data from the Singapore Chinese
the Singapore Chinese                                                                                   knee osteoarthritis (KOA) may be confounded or me-
Health Study. RMD Open              Health Study, a prospective cohort study that recruited
                                    63 257 participants aged 45–74 years from 1993 to                   diated by body mass index, the latter being a strong
2021;7:e001602. doi:10.1136/
                                    1998. At baseline, we assessed diet using a validated               risk factor.
rmdopen-2021-001602
                                    165-­item semiquantitative food frequency questionnaire,         ►► Recent finding that dietary fibre may reduce the risk
                                    together with body mass index (BMI) and lifestyle factors.          of KOA in Western populations needs to be replicated
►► Additional supplemental
                                                                                                        in leaner Asian populations.
material is published online only. Incident TKR cases were identified via record linkage with
To view, please visit the journal   nationwide hospital discharge database through 2017.             What does this study add?
online (http://​dx.d​ oi.​org/​10.​ Results There were 2816 cases of incident TKR due                ►► In a relatively lean Asian population, total fibre in-
1136/r​ mdopen-​2021-​001602).      to severe KOA. The total fibre intake at baseline was               take was not associated with the risk of total knee
                                    not associated with the risk of TKR after adjustment for
                                                                                                        replacement (TKR) for severe KOA.
                                    confounders. Among the food sources of fibre, higher
Manuscripts based on work                                                                            ►► Among the food sources of fibre, higher intake of le-
                                    intake of legumes was associated with a lower risk of
previously presented at a                                                                               gumes was associated with a lower risk of TKR in a
                                    TKR in a dose-­dependent manner; compared with those
conference and published as a                                                                           dose-­dependent manner.
conference abstract in EULAR        having the lowest quartile intake, HR (95% CI) was 0.86
2021.                               (0.76,  0.96) for those having the highest quartile intake       How might this impact on clinical practice or
                                    (p for trend=0.004). This association was consistent after       further developments?
Received 28 January 2021            including BMI in the model and homogeneous across                ►► Further studies are needed to replicate this finding
Accepted 13 June 2021               BMI categories. The consumption of other fibre sources,             and also to study biological plausibility underlying
                                    namely grain products, nuts and seeds, soy food, fruits and         the effect of dietary legumes on development of
                                    vegetables, was not associated with the risk of TKR.                osteoarthritis.
                                    Conclusion Intake of legumes, but not total fibre, was
                                    associated with a reduced risk of TKR. Further research is
                                    needed to replicate our findings and to evaluate possible
                                    biological mechanisms that could explain the effect of          of effective and safe pharmacological treat-
                                    dietary legumes on pathogenesis or progression of KOA.          ment,5 and total knee replacement (TKR)
                                                                                                    is often the only option for end-­stage severe
                                                                                                    KOA6 Obesity is the major risk factor for
                                                                                                    KOA, associated with a 2-­fold to over 10-­fold
                                  INTRODUCTION                                                      increase in the risk of KOA.7 8 Apart from the
© Author(s) (or their             Knee osteoarthritis (KOA) is the most                             increase in biomechanical loading, obesity
employer(s)) 2021. Re-­use        prevalent arthritis affecting 10% of the                          may also increase systemic inflammation,
permitted under CC BY-­NC. No
commercial re-­use. See rights
                                  general population and up to 30% of                               which is thought to play an important role in
and permissions. Published        elderly.1 2 KOA causes knee pain, swelling                        the pathoetiologies of KOA.9 Dietary fibre is
by BMJ.                           and stiffness, leading to activity limitation,                    the edible parts of plants that are resistant to
For numbered affiliations see     participation restriction, emotional distur-                      digestion in the small intestine and partially
end of article.                   bance and eventual loss of independence. It                       fermented in the colon. Dietary patterns rich
 Correspondence to                is one of the leading causes of disability world-                 in food sources of fibre, such as vegetables,
 Dr Ying Ying Leung;              wide.3 4 Despite the huge impact on affected                      legumes and fruits, have been associated
​katyccc@​hotmail.​com            individuals and the society, there is a dearth                    with a lower risk of KOA and higher cartilage

                                           Leung YY, et al. RMD Open 2021;7:e001602. doi:10.1136/rmdopen-2021-001602                                         1
RMD Open

                                                                                                                                            RMD Open: first published as 10.1136/rmdopen-2021-001602 on 30 July 2021. Downloaded from http://rmdopen.bmj.com/ on October 11, 2021 by guest. Protected by copyright.
thickness.10 11 Clinical trials have demonstrated the bene-    Cantonese, who originated from the Fujian and Guang-
ficial effect of dietary fibre on weight loss and reducing     dong Provinces in Southern China, respectively.
systemic inflammation.12 13 Thus, dietary fibre may have
biological effects on the preservation of cartilage and        Baseline exposure assessment
reduction in the risk of KOA.                                  Information at baseline was collected by face-­      to-­
                                                                                                                       face
   Recently, data from the Osteoarthritis Initiative (OAI)     interviews using structured questionnaires.17 Informa-
of 4796 participants and from the Framingham Offspring         tion collected included age at recruitment, sex, educa-
Osteoarthritis Study of 1268 persons in the USA showed         tion level, alcohol consumption, cigarette smoking,
that dietary fibre was associated with a lower risk of inci-   habitual physical activity, hours of sitting, sleeping
dent symptomatic KOA after adjustment for body mass            hours and comorbidities. Habitual dietary intake of the
index (BMI).14 Specifically, in the OAI study, the mean        past year was collected using a validated 165-­item semi-
BMI of those in the lowest quartile intake of fibre (29.4      quantitative Food Frequency Questionnaire (FFQ) that
kg/m2) was 1.8 kg/m2 higher than those in the highest          covered food in the following categories: rice, noodles,
quartile (27.6 kg/m2).14 BMI is a well-­         established   bread and cereals, meats (red meat comprising pork,
major risk factor of KOA,7 8 15 and while overweight or        beef and mutton; poultry comprising chicken and duck;
obesity obviously increases biomechanical loading onto         and seafood comprising fish and shellfish), soy foods,
weight-­bearing joints, chronic inflammation associated        legumes, nuts and seeds, fruits, vegetables, dairy prod-
with obesity has also been hypothesised to be an addi-         ucts, beverages, condiments and preserved foods. For
tional mechanism for the biological plausibility linking       major food items, participants were asked to report their
increased BMI to the development of KOA.16 As those            habitual intake with eight food-­    frequency categories
with higher fibre intake had lower body weight in the OAI      (from never or hardly ever to two or more times a day)
study, the authors later reported that after controlling for   and three portion sizes that were illustrated using actual
BMI, this inverse association between fibre and incident       eating utensils and accompanying photographs. Daily
symptomatic KOA, though still reaching statistical signif-     nutrient and energy intakes were computed from the
icance, was attenuated. Hence, they concluded that this        Singapore Food Composition Table that was developed
inverse association was partially mediated by BMI.             specifically for this cohort.17 The FFQ was subsequently
   As diet is integrally related to BMI, the association       validated in a subset of 810 participants by performing a
between diet and KOA may indeed be confounded or               repeat administration of the FFQ and comparing results
mediated by BMI in observational studies conducted             from two 24-­hour recalls with their results from the FFQ.
in populations with a wide BMI range. As BMI is such           The validation study using these two methods showed
a strong risk factor of KOA, it is difficult to eliminate      similar distributions, with the majority of the mean values
residual confounding effect attributed by adjusting for        of energy and nutrients obtained from the FFQ and from
BMI with statistical modelling alone, and this is partic-      the 24-­hour recalls falling within 10% of each other.
ularly problematic when there is a wide disparity in BMI       For energy and nutrients, the correlation coefficients
among the ordinal categories of dietary intake.                between these two methods ranged from 0.24 to 0.79,17
   In this study, we evaluated the association between fibre   which were comparable to those from previous validation
intake and the risk of TKR due to severe KOA in a rela-        studies in diverse populations.18
tively lean population nested in the Singapore Chinese            In this Chinese cohort in an Asian country, the main
Health Study. In a leaner population with a lower vari-        contributors of fibre were grain products (30.6%),
                                                               fruits and related juices (27.6%), vegetables and related
ability in BMI, differences in the mean BMI by dietary
                                                               juices (21.0%), soy (3.4%), nuts and seeds (1.2%) and
intake will be comparatively smaller, thereby reducing the
                                                               legumes (0.8%), which accounted for 84.6% of the
confounding effect of BMI on the diet–KOA risk associa-
                                                               total dietary fibre. The grain products included noodles
tion. In addition, we evaluated the individual association
                                                               and pasta, rice, bread and pancakes, breakfast cereals,
between specific food sources of fibre in this population,
                                                               biscuits and crackers. Compared with the percentage
namely grain products, legumes, nuts and seeds, soy
                                                               contribution from food sources of dietary fibre in three
food, fruits and vegetables, and the risk of TKR.
                                                               US-­based studies,14 19 the contributions from grains, nuts
                                                               and legumes were considerably lower in this Singapore
                                                               Chinese cohort. Instead, contributions from fruits, vege-
RESEARCH DESIGN AND METHODS                                    tables and other miscellaneous food sources were higher
Study population                                               in our cohort. In the validation study, the mean daily
The Singapore Chinese Health Study is a prospective            intakes of total fibre assessed by 24-­hour recalls and the
cohort of 63 257 Singaporean Chinese, aged 45–74 years         FFQ were 12.2 g and 12.5 g, respectively. The correlation
during recruitment between 1993 and 1998 in Singapore.         coefficients between FFQ-­based and 24-­hour recall-­based
Participants were recruited from public housing estates,       intakes for dietary fibre (kcal-­adjusted) were 0.72, 0.67,
where 86% of Singapore’s population lived during the           0.66 and 0.65 among Cantonese men, Cantonese women,
time of recruitment. Study participants were restricted to     Hokkien men and Hokkien women, respectively. Further-
two major dialect groups in Singapore, the Hokkien and         more, the intake of fibre in this cohort was comparable to

2                                                               Leung YY, et al. RMD Open 2021;7:e001602. doi:10.1136/rmdopen-2021-001602
Osteoarthritis

                                                                                                                                             RMD Open: first published as 10.1136/rmdopen-2021-001602 on 30 July 2021. Downloaded from http://rmdopen.bmj.com/ on October 11, 2021 by guest. Protected by copyright.
the intake level of the participants in the Third National                  who had undergone TKR prior to recruitment, we identi-
Health and Nutrition Examination Survey data in the                         fied 2816 cases of incident TKR due to severe KOA in this
USA.17                                                                      cohort. Deaths in this cohort were ascertained through
    BMI was calculated from weight and height, which                        record linkage analysis with the nationwide Singapore
were self-­reported at baseline interview. A total of 10                    Registry of Births and Deaths. As of 31 December 2011,
349 cohort subjects (16%) did not report either weight                      only 47 (0.07%) subjects from this cohort were known to
or height. For participants with either missing weight                      be lost to follow-­up due to migration out of Singapore or
or height, missing values were imputed using this equa-                     for other reasons ascertained from our follow-­up inter-
tion: weight=y-­intercept+gradient×height. The values for                   views in this cohort.
y-­intercept and gradient were obtained from sex-­specific
weight–height regression lines drawn from all partici-                      Statistical analysis
pants in the cohort with known weights and heights. If                      We calculated person-­   years for each participant from
both weight and height were missing, the missing height                     date of recruitment until TKR, time of death, lost-­to-­
was assigned as the sex-­specific median and the missing                    follow-­up or 31 December 2017, whichever came first.
weight was calculated from the equation.20                                  We used the residual method to derive energy-­adjusted
                                                                            intake of food and fibre for each participant.24 We used
Identification of incident cases of severe KOA                              Cox proportional hazards regression to estimate the HRs
We first identified TKR cases in this cohort and only                       and 95% CIs for the association of quartiles of total fibre
included those that were first-­     occurrence as incident                 intake, as well as food sources of fibre, with the risk of
cases. We then used the diagnosis codes to include                          TKR. Proportional hazards assumption was tested using
primary KOA and exclude cases with diagnosis codes that                     Schoenfeld residuals test, and no violation was observed.
indicated or suggested that the TKR was done for other                      The linear trend was tested by assigning participants
diseases or injuries of the knee joint. We identified partic-               the median value of the quartile they belonged to and
ipants who underwent TKR due to severe KOA via record                       treating this as a continuous variable in the model.24 25 As
linkage with the nationwide MediClaim System hospital                       total energy intake is often related to disease risk because
discharge database.21 22 Although it is used for subse-                     of their associations with body size and physical activity,
quent financial claims, and hence its name, this data-                      we further adjusted for total energy intake as a covar-
base, regardless of the claim status, has captured surgical                 iate in a multivariate nutrient density model to address
procedures and up to three diagnoses per patient for                        the potential problem of confounding by total energy
hospital admissions in all public and private hospitals in                  intake in the nutrient density approach.24 In model 1,
Singapore since 1990. The Singapore Ministry of Health                      we adjusted for sex, age at interview, years of interview
conducts regular check on the MediClaims Database to                        (1993–1995, 1996–1998), total energy intake (kcal/day),
ensure its completeness.23 We identified cases of TKR in                    dialect (Hokkien, Cantonese) and educational level
this cohort using the operation codes SB010K, SB012K                        (none, primary, secondary and more). In model 2, we
and SB013K (unilateral right, unilateral left and bilateral                 further adjusted for BMI (kg/m2), total physical activity
TKR) for linkage up to December 2011 and operation                          duration (hours/week), sleep duration (hours/day) and
codes SB716K and SB809K (TKR unilateral and TKR bilat-                      sitting duration (hours/day), smoking status (never, past,
eral) for linkage from January 2012 to December 2017.                       current) and a history of hypertension, diabetes, heart
We only included information for the first TKR in cases                     attack and stroke (yes/no). These covariates were chosen
with more than one episode. Among the TKR cases, we                         because they either are known risk factors of KOA in the
further included only cases with a diagnosis of KOA using                   literature15 or have been reported to affect the risk of
International Classification of Diseases (ICD)-9 code 715                   TKR in our cohort.8 21 26 27 Finally, to investigate the asso-
and all subcategories for linkage up to December 2011,                      ciation between individual food sources of fibre and the
and ICD-10 code M17 and all subcategories for linkage                       risk of TKR, we added a model 3 that included mutual
from January 2012 to December 2017. We excluded all                         adjustment for all the other food sources.
cases with any mention of diagnoses that included septic                       We tested for interaction between dietary variables
arthritis, osteomyelitis, villonodular synovitis, rheuma-                   and BMI or sex by adding their cross-­product terms into
toid arthritis, psoriatic arthritis, ankylosing spondylitis                 regression models. For BMI, we used the cut-­point of 23
and other inflammatory arthritis, or a secondary cause                      kg/m2 to divide participants into lean and heavy catego-
of KOA, such as avascular or aseptic necrosis of joint,                     ries. This cut-­point was recommended for the definition
meniscus or ligament injuries, old tear or rupture or                       of cardiovascular risk for Asians by the WHO.28 Finally,
injuries, delay in development, other acquired deform-                      sensitivity analysis was conducted by excluding partic-
ities of knee and neuritis. Although we did not ask for                     ipants with imputed BMI (10 349 participants), those
the history of knee injury in our study, we excluded TKR                    with an extreme energy intake of 3000 kcal/
cases with osteoarthritis secondary to other underlying                     day for women (n=584) and 3700 kcal/day for
conditions, including severe injury, if this was mentioned                  men (n=476), and those with a follow-­up time of less
in the medical records, so that we could focus on TKR                       than 5 years (333 TKR cases and 3495 non-­cases). We
done for primary KOA. After excluding 128 participants                      also conducted an analysis that only included cases with

Leung YY, et al. RMD Open 2021;7:e001602. doi:10.1136/rmdopen-2021-001602                                                              3
RMD Open

                                                                                                                                                  RMD Open: first published as 10.1136/rmdopen-2021-001602 on 30 July 2021. Downloaded from http://rmdopen.bmj.com/ on October 11, 2021 by guest. Protected by copyright.
                                                                   without the inclusion of BMI or comorbidities. We also
                                                                   used the method described by Lange et al29 in mediation
                                                                   analysis to estimate the direct effects in the association
                                                                   between dietary fibre/legumes and the risk of TKR, as well
                                                                   as the indirect effects mediated by BMI. We conducted
                                                                   all statistical analyses using SAS V.9.4 (SAS Institute). All
                                                                   reported p values were two-­sided, and p
Osteoarthritis

                                                                                                                                                       RMD Open: first published as 10.1136/rmdopen-2021-001602 on 30 July 2021. Downloaded from http://rmdopen.bmj.com/ on October 11, 2021 by guest. Protected by copyright.
 Table 2 Baseline characteristics of cohort participants stratified by quartiles of total fibre intake: the Singapore Chinese
 Health Study (1993–2017)
                                                    Quartile of total fibre intake
 Characteristics                                    Q1                      Q2                         Q3                      Q4
 Number of participants                             15 796                  15 775                     15 785                  15 773
 Number of TKR cases                                528                     712                        767                     809
 Women, n (%)                                       5923 (37.5)             9241 (58.6)                9892 (62.7)             10 135 (64.3)
 Age, years*                                        56.6 (8.0)              56.8 (8.1)                 56.5 (8.0)              56.0 (7.9)
 Body mass index, kg/m*†                            23.0 (3.3)              23.2 (3.3)                 23.2 (3.2)              23.2 (3.3)
 Dialect group, n (%)
  Hokkien                                          8812 (55.8)             8760 (55.5)                8314 (52.7)             8027 (50.9)
  Cantonese                                        6984 (44.2)             7015 (44.5)                7471 (47.3)             7746 (49.1)
 Secondary school or higher, n (%)                  3610 (22.9)             3770 (23.9)                4605 (29.2)             5881 (37.3)
 Ever smoker, n (%)                                 7634 (48.3)             4769 (30.2)                3805 (24.1)             3107 (19.7)
 Hypertension, n (%)                                3227 (20.4)             3799 (24.1)                3881 (24.6)             4077 (25.9)
 Diabetes, n (%)                                    1150 (7.3)              1429 (9.1)                 1511 (9.6)              1581 (10.0)
 Coronary heart disease, n (%)                      512 (3.2)               624 (4.0)                  672 (4.3)               780 (5.0)
 Stroke, n (%)                                      207 (1.3)               223 (1.4)                  229 (1.5)               283 (1.8)
 Physical activity, hour/week*                      1.9±5.1                 1.3±3.8                    1.4±3.8                 1.8±3.9
 Sitting, hours/day*                                6.6±3.8                 6.3±3.5                    6.4±3.5                 6.7±3.6
 Sleep, hours/day*                                  7.04±1.2                7.03±1.1                   6.99±1.1                6.98±1.1
 Total energy intake, kcal/day*‡                    1,697.3±622.9           1,411.3±524.6              1,462.0±497.3           1,646.5±553.3
 Total fibre, g/day*‡                               9.1±4.0                 10.1±3.9                   12.8±3.7                18.8±5.8
 Grain products, g/day*†‡                           649.7±254.6             497.9±198.7                485.5±191.5             507.3±203.8
 Legumes, g/day*‡                                   2.5±3.8                 2.7±3.9                    3.2±4.7                 4.2±6.1
 Nuts and seeds, g/day*‡                            2.4±3.7                 2.3±3.5                    2.5±3.8                 3.1±4.9
 Soy food, g/day*‡                                  90.7±76.8               95.9±75.6                  112.4±86.0              147.8±119.4
 Fruits, g/day*‡                                    97.3±89.0               139.3±95.9                 207.5±113.1             366.0±207.0
 Vegetables, g/day*‡                                86.7±46.6               95.2±47.8                  112.1±54.3              148.4±80.8
                                    2
 Statistical testing based on the χ test for categorical variables and analysis of variance for continuous variables; all p values
RMD Open

                                                                                                                                                           RMD Open: first published as 10.1136/rmdopen-2021-001602 on 30 July 2021. Downloaded from http://rmdopen.bmj.com/ on October 11, 2021 by guest. Protected by copyright.
Table 3 HRs (95% CIs) for risk of total knee replacement according to intake of fibre: the Singapore Chinese Health Study
(1993–2017)
Quartiles of energy-­adjusted food
intake                             Q1                        Q2                        Q3                      Q4                      P for trend*
Total fibre
 Median, g/day†                       8.55                  9.43                      12.19                   17.70
 Cases/ person-­years                 528/2 75 050          712/280 456               767/284 394             809/287 955
 Multivariate model 1                 1.00                  1.08 (0.96 to 1.21)       1.12 (1.00 to 1.26)     1.18 (1.05 to 1.32)      0.003
 Multivariate model 2                 1.00                  1.06 (0.95 to 1.19)       1.09 (0.97 to 1.22)     1.11 (0.99 to 1.24)      0.088
Grain products‡
 Median, g/day†                       374.45                420.21                    507.04                  793.28
 Cases/person-­years                  744/289 413           808/283 836               684/280 047             580/274 558
 Multivariate model 1                 1.00                  1.07 (0.96 to 1.18)       0.93 (0.83 to 1.03)     0.91 (0.82 to 1.02)      0.024
 Multivariate model 2                 1.00                  1.09 (0.98 to 1.21)       0.96 (0.86 to 1.06)     0.95 (0.85 to 1.06)      0.110
 Multivariate model 3                 1.00                  1.09 (0.98 to 1.21)       0.95 (0.85 to 1.06)     0.94 (0.82 to 1.06)      0.120
Legumes
 Median, g/day†                       0.11                  0.43                      2.11                    7.01
 Cases/ person-­years                 638/ 80 170           754/278 001               756/280 407             668/289 275
 Multivariate model 1                 1.00                  0.97 (0.87 to 1.09)       0.95 (0.84 to 1.06)     0.84 (0.76 to 0.94)
Osteoarthritis

                                                                                                                                           RMD Open: first published as 10.1136/rmdopen-2021-001602 on 30 July 2021. Downloaded from http://rmdopen.bmj.com/ on October 11, 2021 by guest. Protected by copyright.
the association was no longer statistically significant after                  The strength of our study is that the BMI levels across
adjustment with other confounders and food sources of                       quartiles of fibre intake were quite homogeneous
fibre. The other sources of fibre, namely grain products,                   among the relatively leaner participants in this prospec-
soy food, fruits and vegetables, were not significantly asso-               tive cohort study in Singapore. Hence, expectantly,
ciated with the risk of TKR.                                                the associations between total fibre or food intake and
   In the analysis stratified by BMI or sex, the intake of                  risk of TKR were essentially unchanged in the models
total fibre was not associated with the risk of TKR in any                  with and without BMI adjustment, thus indicating that
stratum, and the risk estimates were not dissimilar by sex                  any mediation or confounding effect from BMI was
or BMI categories (p for interaction ≥0.32). Conversely,                    minimal. Another strength of our study is its prospective
the inverse association between the intake of legumes                       population-­based cohort design with a large sample size
and the risk of TKR remained statistically significant in                   over a long follow-­up time. TKR cases were identified via
women and also in those with a BMI ≥23 kg/m2. Although                      linkage with a comprehensive nationwide database that
the inverse associations in men and in those with BMI
RMD Open

                                                                                                                                                                       RMD Open: first published as 10.1136/rmdopen-2021-001602 on 30 July 2021. Downloaded from http://rmdopen.bmj.com/ on October 11, 2021 by guest. Protected by copyright.
valid for epidemiological studies across different popu-                                ORCID iD
                                                                                        Ying Ying Leung http://​orcid.​org/​0000-​0001-​8492-​6342
lations.39 Third, diet and BMI were only assessed once at
baseline; therefore, we lacked information about poten-
tial changes in these exposures that could have occurred
during the follow-­  up period. Although FFQs tend to
                                                                                        REFERENCES
overestimate and underestimate the absolute intake of                                    1 Wallace IJ, Worthington S, Felson DT, et al. Knee osteoarthritis has
various nutrients and foods, for most nutrient and foods,                                  doubled in prevalence since the mid-­20th century. Proc Natl Acad
                                                                                           Sci U S A 2017;114:9332–6.
the ability of the FFQ to rank subjects according to their                               2 Leung Y-­Y, Ma S, Noviani M, et al. Validation of screening
dietary intake has been shown to be acceptable to good,                                    questionnaires for evaluation of knee osteoarthritis prevalence in the
and this is the basis of all epidemiological studies that                                  general population of Singapore. Int J Rheum Dis 2018;21:629–38.
                                                                                         3 Vos T, Flaxman AD, Naghavi M, et al. Years lived with disability
have used FFQ to study diet–disease relationships.40 There                                 (YLDs) for 1160 sequelae of 289 diseases and injuries 1990-2010:
was a 64-­fold difference in the median intake of legumes                                  a systematic analysis for the global burden of disease study 2010.
                                                                                           Lancet 2012;380:2163–96.
between the lowest and the highest quartile in the Singa-                                4 The burden of disease in Singapore, 1990–2017: an overview of the
pore Chinese Health Study cohort that was sufficient to                                    global burden of disease study 2017 results, 2019. Available: https://
demonstrate the inverse legume–TKR association in a                                        www.​moh.​gov.​sg/​docs/​librariesprovider5/​default-​document-​library/​
                                                                                           gbd_​2017_​singapore_​repo​rtce​6bb0​b3ad​1a49​c19e​e6eb​adc1​273b18.​
dose-­dependent manner. Finally, residual confounding                                      pdf [Accessed 10 Jun 2021].
cannot be completely ruled out in our study due to the                                   5 Bannuru RR, Osani MC, Vaysbrot EE, et al. OARSI guidelines
                                                                                           for the non-­surgical management of knee, hip, and polyarticular
limitation of the observational design.                                                    osteoarthritis. Osteoarthritis Cartilage 2019;27:1578–89.
   In conclusion, in this prospective cohort study, we                                   6 NIH Consensus Panel. Nih consensus statement on total knee
                                                                                           replacement December 8-10, 2003. J Bone Joint Surg Am
provided novel epidemiology evidence that although                                         2004;86:1328–35.
total fibre intake was not associated with severe KOA,                                   7 Jiang L, Tian W, Wang Y, et al. Body mass index and susceptibility
increased intake of legumes could possibly reduce the                                      to knee osteoarthritis: a systematic review and meta-­analysis. Joint
                                                                                           Bone Spine 2012;79:291–7.
risk of severe KOA in a dose-­dependent manner. Further                                  8 Leung Y-­Y, Allen JC, Noviani M, et al. Association between body
research is needed to replicate our findings, and the                                      mass index and risk of total knee replacement, the Singapore
                                                                                           Chinese Health study. Osteoarthritis Cartilage 2015;23:41–7.
biological mechanisms underlying the effect of dietary                                   9 Berenbaum F. Osteoarthritis as an inflammatory disease
legumes on pathogenesis or progression of osteoarthritis                                   (osteoarthritis is not osteoarthrosis!). Osteoarthritis and Cartilage
warrant further investigations.                                                            2013;21:16–21.
                                                                                        10 Veronese N, La Tegola L, Crepaldi G, et al. The association between
                                                                                           the Mediterranean diet and magnetic resonance parameters for knee
Author affiliations                                                                        osteoarthritis: data from the osteoarthritis initiative. Clin Rheumatol
1
 Rheumatology & Immunology, Singapore General Hospital, Singapore                          2018;37:2187–93.
2
 Duke-­NUS Medical School, Singapore                                                    11 Veronese N, Koyanagi A, Stubbs B, et al. Mediterranean diet and
3
 Healthy Longevity Translational Research Programme, Yong Loo Lin School of                knee osteoarthritis outcomes: a longitudinal cohort study. Clinical
                                                                                           Nutrition 2019;38:2735–9.
Medicine, National University of Singapore, Singapore                                   12 Jovanovski E, Mazhar N, Komishon A, et al. Can dietary viscous
4
 Government of Singapore Ministry of Health, Singapore                                     fiber affect body weight independently of an energy-­restrictive diet?
5
 National Public Health and Epidemiology Unit, National Centre for Infectious              A systematic review and meta-­analysis of randomized controlled
Diseases, Singapore                                                                        trials. Am J Clin Nutr 2020;111:471–85.
6
 Division of Cancer Control and Population Sciences, UPMC Hillman Cancer Centre,        13 Jiao J, Xu J-­Y, Zhang W, et al. Effect of dietary fiber on circulating
                                                                                           C-­reactive protein in overweight and obese adults: a meta-­analysis
University of Pittsburgh, Pittsburgh, Pennsylvania, USA
7                                                                                          of randomized controlled trials. Int J Food Sci Nutr 2015;66:114–9.
 Department of Epidemiology, University of Pittsburgh, Pittsburgh, Pennsylvania,        14 Dai Z, Niu J, Zhang Y, et al. Dietary intake of fibre and risk of
USA                                                                                        knee osteoarthritis in two us prospective cohorts. Ann Rheum Dis
                                                                                           2017;76:1411–9.
Contributors YYL and WPK designed and conducted the research; WKP, JMY, LWA             15 Blagojevic M, Jinks C, Jeffery A, et al. Risk factors for onset of
                                                                                           osteoarthritis of the knee in older adults: a systematic review and
and KBT collected the data; AJ and WPK analysed data; YYL and WPK wrote the
                                                                                           meta-­analysis. Osteoarthritis and Cartilage 2010;18:24–33.
paper; and WPK had primary responsibility for final content. All authors critically     16 Thijssen E, van Caam A, van der Kraan PM. Obesity and
reviewed and approved the final manuscript.                                                osteoarthritis, more than just wear and tear: pivotal roles for inflamed
Funding This study was supported by the National Institutes of Health, USA (NIH            adipose tissue and dyslipidaemia in obesity-­induced osteoarthritis.
R01 CA144034 and UM1 CA182876). YYL and WKP were supported by the National                 Rheumatology 2015;54:588–600.
Medical Research Council, Singapore (NMRC/CSA-­Inv/0022/2017 and MOH-­                  17 Hankin JH, Stram DO, Arakawa K, et al. Singapore Chinese Health
CSASI19nov-0001, respectively).                                                            study: development, validation, and calibration of the quantitative
                                                                                           food frequency questionnaire. Nutr Cancer 2001;39:187–95.
Competing interests None declared.                                                      18 McDowell MA, Briefel RR, Alaimo K. Energy and macronutrient
Patient consent for publication Not required.                                              intakes of persons ages 2 months and over in the United States:
                                                                                           third National health and nutrition examination survey, phase 1,
Ethics approval The Institutional Review Board of the National University of               1988-91. Adv Data 2018:1–24.
Singapore read and approved the study protocol. All participants signed informed        19 McGill CR, Fulgoni VL, Devareddy L. Ten-­Year trends in fiber
consents prior to recruitment.                                                             and whole grain intakes and food sources for the United States
                                                                                           population: National health and nutrition examination survey 2001-
Provenance and peer review Not commissioned; externally peer reviewed.                     2010. Nutrients 2015;7:1119–30.
Data availability statement Data are available upon reasonable request. All data        20 Koh W-­P, Yuan J-­M, Wang R, et al. Body mass index and smoking-­
are available upon reasonable request with email to corresponding author.                  related lung cancer risk in the Singapore Chinese Health study. Br J
                                                                                           Cancer 2010;102:610–4.
Open access This is an open access article distributed in accordance with the           21 Leung Y-­Y, Ang L-­W, Thumboo J, et al. Cigarette smoking and risk
Creative Commons Attribution Non Commercial (CC BY-­NC 4.0) license, which                 of total knee replacement for severe osteoarthritis among Chinese
permits others to distribute, remix, adapt, build upon this work non-­commercially,        in Singapore – the Singapore Chinese Health study. Osteoarthritis
and license their derivative works on different terms, provided the original work is       Cartilage 2014;22:764–70.
properly cited, appropriate credit is given, any changes made indicated, and the        22 Leung Y-­Y, Talaei M, Ang L-­W, et al. Reproductive factors and
use is non-­commercial. See: http://​creativecommons.​org/l​ icenses/​by-n​ c/​4.​0/.      risk of total knee replacement due to severe knee osteoarthritis in

8                                                                                        Leung YY, et al. RMD Open 2021;7:e001602. doi:10.1136/rmdopen-2021-001602
Osteoarthritis

                                                                                                                                                            RMD Open: first published as 10.1136/rmdopen-2021-001602 on 30 July 2021. Downloaded from http://rmdopen.bmj.com/ on October 11, 2021 by guest. Protected by copyright.
     women, the Singapore Chinese Health study. Osteoarthritis Cartilage        32 Bouchenak M, Lamri-­Senhadji M. Nutritional quality of legumes, and
     2019;27:1129–37.                                                              their role in cardiometabolic risk prevention: a review. J Med Food
23   Leung Y-­Y, Ang L-­W, Allen JC, et al. Response to letters to the             2013;16:185–98.
     editors: 1. more details on the database used by the study should be       33 Abeysekara S, Chilibeck PD, Vatanparast H, et al. A pulse-­based
     provided. 2. Mediclaim hospital discharge system and income levels            diet is effective for reducing total and LDL-­cholesterol in older
     of cohort. Osteoarthritis Cartilage 2015;23:499–500.                          adults. Br J Nutr 2012;108 Suppl 1:S103–10.
24   Willett WC, Howe GR, Kushi LH. Adjustment for total energy                 34 Abete I, Parra D, Martinez JA. Legume-, fish-, or high-­protein-­based
     intake in epidemiologic studies. Am J Clin Nutr 1997;65:1220S–8.
                                                                                   hypocaloric diets: effects on weight loss and mitochondrial oxidation
     discussion 9S-­31S.
25   Brown CC, Kipnis V, Freedman LS, et al. Energy adjustment                     in obese men. J Med Food 2009;12:100–8.
     methods for nutritional epidemiology: the effect of categorization.        35 Morquette B, Shi Q, Lavigne P, et al. Production of lipid
     Am J Epidemiol 1994;139:323–38.                                               peroxidation products in osteoarthritic tissues: new evidence
26   Leung Y-­Y, Allen JC, Ang L-­W, et al. Diabetes mellitus and the risk of      linking 4-­hydroxynonenal to cartilage degradation. Arthritis Rheum
     total knee replacement among Chinese in Singapore, the Singapore              2006;54:271–81.
     Chinese Health study. Sci Rep 2017;7:40671.                                36 Blanco FJ, Rego-­Pérez I. Mitochondria and mitophagy: biosensors
27   Leung YY, Bin Abd Razak HR, Talaei M, et al. Duration of physical             for cartilage degradation and osteoarthritis. Osteoarthritis Cartilage
     activity, sitting, sleep and the risk of total knee replacement among         2018;26:989–91.
     Chinese in Singapore, the Singapore Chinese Health study. PLoS             37 Collins JA, Diekman BO, Loeser RF. Targeting aging for
     One 2018;13:e0202554.                                                         disease modification in osteoarthritis. Curr Opin Rheumatol
28   WHO Expert Consultation. Appropriate body-­mass index for                     2018;30:101–7.
     Asian populations and its implications for policy and intervention         38 McAlindon TE, Driban JB, Henrotin Y, et al. OARSI clinical trials
     strategies. Lancet 2004;363:157–63.
                                                                                   recommendations: design, conduct, and reporting of clinical trials for
29   Lange T, Vansteelandt S, Bekaert M. A simple unified approach
                                                                                   knee osteoarthritis. Osteoarthritis Cartilage 2015;23:747–60.
     for estimating natural direct and indirect effects. Am J Epidemiol
     2012;176:190–5.                                                            39 Connor Gorber S, Tremblay M, Moher D, et al. A comparison of
30   Yao Y, Cheng X, Wang L, et al. Biological potential of sixteen                direct vs. self-­report measures for assessing height, weight and
     legumes in China. Int J Mol Sci 2011;12:7048–58.                              body mass index: a systematic review. Obes Rev 2007;8:307–26.
31   Singh B, Singh JP, Kaur A, et al. Phenolic composition and                 40 Streppel MT, de Vries JHM, Meijboom S, et al. Relative validity of the
     antioxidant potential of grain legume seeds: a review. Food Res Int           food frequency questionnaire used to assess dietary intake in the
     2017;101:1–16.                                                                Leiden longevity study. Nutr J 2013;12:75.

Leung YY, et al. RMD Open 2021;7:e001602. doi:10.1136/rmdopen-2021-001602                                                                              9
You can also read